Investigating the Relationship Between Debiasing and Artifact Removal using Saliency Maps

Lukasz Sztukiewicz, Ignacy Stępka, Michał Wiliński, Jerzy Stefanowski
2/28/2025
cs.LGcs.AIcs.CY

No official code found. Best implementation: DetoxAI/saliency-fairness-metrics (2 stars, MIT, last commit 2025-05-15).

Abstract

The widespread adoption of machine learning systems has raised critical concerns about fairness and bias, making mitigating harmful biases essential for AI development. In this paper, we investigate the relationship between debiasing and removing artifacts in neural networks for computer vision tasks. First, we introduce a set of novel XAI-based metrics that analyze saliency maps to assess shifts in a model's decision-making process. Then, we demonstrate that successful debiasing methods systematically redirect model focus away from protected attributes. Finally, we show that techniques originally developed for artifact removal can be effectively repurposed for improving fairness. These findings provide evidence for the existence of a bidirectional connection between ensuring fairness and removing artifacts corresponding to protected attributes.

View on arXivView PDF

Code Implementations(1)

DetoxAI/saliency-fairness-metricsCommunity59%
MIT

This is a repository containing code for "Investigating the Relationship Between Debiasing and Artifact Removal using Saliency Maps" paper at XAI-2025

20PythonApr 23, 20251 years agoMIT

Similar Papers

Interpretable Debiasing of Vision-Language Models for Social Fairness

86% match2026

DetoxAI: a Python Toolkit for Debiasing Deep Learning Models in Computer Vision

83% match2025

Saliency Methods are Encoders: Analysing Logical Relations Towards Interpretation

82% match2024

Evaluating the Fairness of Discriminative Foundation Models in Computer Vision

82% match1 repo2023

A matter of attitude: Focusing on positive and active gradients to boost saliency maps

82% match1 repo12023

Bridging the Gap Between Saliency Prediction and Image Quality Assessment

81% match2024

Cite this paper

@article{sztukiewicz2025investigating,
  title  = {Investigating the Relationship Between Debiasing and Artifact Removal using Saliency Maps},
  author = {Lukasz Sztukiewicz and Ignacy Stępka and Michał Wiliński and Jerzy Stefanowski},
  year   = {2025},
  eprint = {2503.00234},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2503.00234v3}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2503.00234.svg)](https://codeofpaper.com/paper/2503.00234)

Discussion